Planar quadrilateral (PQ) meshes play an important role in Architectural Geometry, with ongoing research focused on developing effective tools for their design. Cone-nets are a special class of regular PQ meshes in which one family of PQ strips forms discrete (projective) cones. This paper builds on recent advances in the study of cone-nets and presents a constructive implementation in an interactive design tool that enables intuitive real-time exploration of the design space of cone-nets within the Grasshopper/Rhino environment. We provide novel theoretical insights into cone-nets, introduce a user-friendly interface, and incorporate a correction mechanism to repair degenerate configurations. Our aim is to make cone-nets more accessible to the broader community of practitioners and to promote their application in advanced architectural design scenarios.

Using (2712.4 ± 14.3) × 106 ψ(3686) candidates collected by the BESIII detector operating at the BEPCII storage ring, the decays $ {\chi}{cJ}\left(J=0,1,2\right)\to p\overline{p}\eta \eta $ are observed for the first time through the radiative transition ψ(3686) → γχ cJ . The statistical significances for χ cJ signals are all larger than 5σ. The branching fractions of $ {\chi}{c0,1,2}\to p\overline{p}\eta \eta $ are determined to be (5.75 ± 0.59 ± 0.42) × 10 −5, (1.40 ± 0.33 ± 0.17) × 10 −5, and (2.64 ± 0.40 ± 0.27) × 10 −5, respectively, where the first uncertainties are statistical and the second systematic. No evident resonant structures are found in the $ p\overline{p} $ and $ p\eta /\overline{p}\eta $ systems.

Anionic Ln(iv) complexes were synthesised by tuning the reaction condition, demonstrating the possibility of accessing charged Pr(iv) complexes as a tool to manipulate the redox potential and therefore lead to more stable complexes.

Efficient traffic monitoring is crucial for managing urban transportation networks, especially under congested and dynamically changing traffic conditions. Drones offer a scalable and cost-effective alternative to fixed sensor networks. However, deploying fleets of low-cost drones for traffic monitoring poses challenges in adaptability, scalability, and real-time operation. To address these issues, we propose a learning-based framework for decentralized traffic monitoring with drone swarms, targeting the uneven and unpredictable distribution of monitoring needs across urban areas. Our approach introduces a semi-decentralized reinforcement learning model, which trains a single Q-function using the collective experience of the swarm. This model supports full scalability, flexible deployment, and, when hardware allows, the online adaptation of each drone’s action-selection mechanism. We first train and evaluate the model in a synthetic traffic environment, followed by a case study using real traffic data from Shenzhen, China, to validate its performance and demonstrate its potential for real-world applications in complex urban monitoring tasks.

This talk demonstrates the use of CARMA (=karma for cars) as a fair solution to the morning commute congestion. We consider heterogeneous commuters traveling through a single bottleneck that differ in the value of time (VOT), generalizing the notion of VOT to vary dynamically on each day (e.g., according to trip purpose and urgency) rather than being a static characteristic of each individual. In our CARMA scheme, the bottleneck is divided into a fast lane that is kept in free flow and a slow lane that is subject to congestion. Commuters use karma to bid for access to the fast lane, and those who get outbid or do not participate in the scheme instead use the slow lane. At the end of each day, karma collected from the bidders is redistributed, and the process repeats day by day. We specialize the karma economy mean-field game model to this setting and analyze pthe roperties of its mean-field equilibrium. Unlike existing monetary schemes, CARMA is demonstrated to achieve (a) an equitable traffic assignment with respect to heterogeneous income classes and (b) a strong Pareto improvement in the long-term average travel disutility with respect to no policy intervention. Moreover, CARMA can retain the same congestion reduction as an optimal monetary tolling scheme under uniform karma redistribution and even outperforms tolling under a well-designed redistribution scheme.

Quantum materials exhibit exotic properties and intriguing collective phenomena arising from quantum degrees of freedom and strong many-body correlations, offering significant potential for novel technological applications. Photon-based spectroscopies, including angle-resolved photoemission spectroscopy (ARPES) and resonant inelastic X-ray spectroscopy (RIXS), have become essential tools for investigating these properties both in equilibrium and out-of-equilibrium.

This thesis develops advanced theoretical methods for photon-based spectroscopies. Chapter 2 introduces the Wannier-ARPES formalism, in order to calculate photoemission matrix elements from Wannier-function based slab tight-binding models. A key contribution is a microscopic theory for circular dichroism ARPES (CD-ARPES), enabling the mapping of wavefunction-related properties like orbital textures and band topology. Using Wannier-ARPES formalism, experimental CD-ARPES results can be accurately simulated, clearly distinguishing intra-atomic terms --which reflect local orbital angular momentum (OAM) -- and inter-atomic interference terms, which introduce universal photon-energy dependencies.

Chapter 3 explores dynamics observed through time-resolved ARPES (tr-ARPES), with particular focus on Floquet engineering. The chapter highlights the relation between Floquet states and multiphoton photoemission (mPP) with varying field strengths. In the intermediate field strength, photoemission probes Floquet quasienergy splitting. In a stronger field, complex non-adiabatic dynamics among Floquet states emerge, directly influencing observed mPP features. These insights are critical for realizing Floquet engineering and provide valuable views into nonlinear driven dynamics.

Chapter 4 focuses on resonant inelastic X-ray spectroscopy (RIXS) for strongly correlated systems, where various elementary excitations can be probed. A cluster model approach, combining exact diagonalization and an effective Anderson impurity model derived from \textit{ab-initio} parameters, is used to describe many-body excitations and the RIXS spectrum. The method successfully captures intensity variations of d-d excitations across magnetic phase transitions in YBaCuFeO5.

Overall, this thesis bridges theoretical advancements with photon-based spectroscopy experiments, providing tools not only to interpret complex spectroscopic data but also to guide future experimental studies for deeper insights into quantum materials.

This thesis presents the development and application of novel Neural Network Quantum States (NQS) for the simulation of strongly correlated quantum matter in continuous space. We augment traditional Slater-Jastrow-Backflow ansaetze with modern machine learning architectures, such as permutation-invariant DeepSets and permutation-equivariant graph neural networks (GNNs), dubbed MP-NQS, to construct compact, flexible and highly expressive variational models for both bosonic and fermionic systems.

A key innovation of this work is the incorporation of periodic boundary conditions into the network design of the NQS, allowing the accurate simulation of condensed matter systems and materials and the computation of thermodynamic properties from first principles. We demonstrate the effectiveness of this approach by simulating the phase diagrams of benchmark systems, including $^4$He in one and two dimensions and the homogeneous electron gas in three dimensions, capturing superfluidity, crystallization, and other emergent phenomena.

For fermionic systems, we further enhance the variational ansatz by integrating the Pfaffian determinant as anti-symmetric prior, allowing us to describe pairing correlations in ultracold Fermi gases across the BCSâ BEC crossover. In addition to bulk applications, we extend the framework to molecular systems, demonstrate its applicability to small molecules, and compute real-time dynamical properties using the time-dependent variational principle. Furthermore, we introduce an algorithm to access their finite-temperature properties and nuclear quantum effects through variational (path-integral) molecular dynamics.

Overall, the methods introduced in this work significantly broaden the applicability of variational quantum Monte Carlo by combining physical priors with the flexibility of deep learning. They pave the way for accurate, scalable, and transferable modeling of quantum matter, with potential impact on quantum chemistry, materials science, and strongly correlated condensed matter systems.

Hydrogels are widely used in cell biology and tissue engineering because of their high water content, biocompatibility, and adjustable mechanical properties. These qualities make them ideal for mimicking the extracellular matrix and creating soft devices that interact with biological tissues. However, their lack of electronic conductivity and low mechanical stiffness limit their application in bioelectronics and load-bearing uses such as bone tissue engineering. Efforts to overcome these limitations by adding conductive fillers or biominerals often reduce processability, especially through additive manufacturing techniques like direct laser writing (DLW) or extrusion-based 3D printing. To address these issues, I explore two strategies centered on the bottom-up, in-situ formation of functional fillers within hydrogels. By spatially localizing these fillers, the hydrogel gains electrical or mechanical functions not present in the original material, without losing compatibility with advanced manufacturing methods. In the first strategy, I use two-photon DLW to create high-resolution silver microstructures inside optically clear, soft hydrogel matrices. This technique reduces silver salts within the hydrogel through photoreduction, resulting in conductive features with resolutions as small as 5 µm and conductivities up to 1505 S/cm, without pre-mixing conductive fillers. This separates hydrogel formulation from filler addition and enables the creation of embedded or surface-exposed conductive pathways, opening new possibilities for soft, hydrogel-based bioelectronic devices. The second strategy uses a nature-inspired approach to 3D print biomineralized hydrogel scaffolds. Ureolytic bacteria are encapsulated in printable microgels to create a bioactive ink capable of inducing calcium carbonate mineralization in situ. This mineralization happens after printing, allowing independent optimization of the ink's rheological properties for printability. Spatial and temporal control over biomineralization results in scaffolds with mineral content up to 93% by weight. The microgels act as sacrificial templates, guiding the development of a 3D porous network that mimics trabecular bone architecture and achieves compressive strengths up to 3.5 MPa. This process uses only mild, biocompatible reagents and avoids high-temperature sintering. I demonstrate proof-of-concept applications in bone tissue engineering by printing complex porous structures and suggest potential use in art restoration. Together, these methods demonstrate that in-situ formation of functional fillers allows the spatial integration of conductivity and stiffness into hydrogels without compromising optical or rheological properties essential for additive manufacturing. I also outline future directions to enhance and expand these approaches, including using DLW to create 3D interconnects inspired by microfabrication techniques and developing hybrid scaffolds that combine electronic and mechanical functionalities for advanced tissue engineering. These innovations establish bottom-up hydrogel functionalization as a versatile platform for next-generation cell culture materials, bioelectronic devices, and engineered tissue scaffolds.

Multi-agent reinforcement learning (MARL) has emerged as a compelling framework for modeling the collaborative behavior of autonomous agents operating in interconnected systems. The potential for agents to collectively achieve goals that are infeasible for any single unit makes MARL a powerful tool across a wide range of domains. However, the multi-agent setting introduces distinct challenges that require specialized solutions. This thesis addresses two fundamental challenges in MARL: (i) effective deep exploration, and (ii) global state estimation under partial observability. To this end, we leverage the networked structure of agents and their communication capabilities to develop decentralized learning algorithms that facilitate robust and scalable collaboration under uncertain conditions.

First, we propose a novel, counting-free deep exploration algorithm for MARL that guarantees all state-action pairs are visited infinitely often. Deep exploration is essential for avoiding suboptimal learning in environments with sparse or deceptively structured rewards. Our method distributes an ensemble of value estimates across the network of agents and uses statistical variance to guide exploration. The count-free nature of the design makes it suitable for large or continuous state spaces. Theoretical guarantees are established for sufficient exploration, and the approach is validated through extensive simulations.

Second, we address the challenge of global state estimation in partially observable environments, where agents have access only to local, incomplete observations. Individually, these observations are insufficient to recover the global state; however, through local communication, agents can collaboratively estimate it. We explore two social learning-based strategies to tackle this issue: standard and adaptive social learning. Standard social learning does not impose constraints on state dynamics but introduces a two-time-scale learning structure. We provide theoretical analysis showing that MARL combined with this approach achieves $\epsilon$-optimality with respect to the fully observable baseline.

To overcome the limitation of two-time-scale learning, we introduce an adaptive social learning method that enables single-time-scale integration of state estimation and reinforcement learning, assuming slowly evolving state dynamics. Under appropriate choices of the adaptation and learning parameters, we show that the proposed method also achieves $\epsilon$-optimal performance. Both methods are fully decentralized, rely solely on local communication, and are supported by rigorous convergence guarantees. Empirical evaluations further confirm the effectiveness of both approaches.

With the growing demand for miniaturized, self-powered devices, energy harvesting technologies that can exploit ambient and physiological energy sources have gained increasing attention. Beyond conventional batteries, strategies to convert energy from the human bodyâ such as heat or mechanical deformationâ into electricity are especially attractive for wearable electronics and distributed sensing platforms. These processes often rely on polar materials such as polyvinylidene fluoride (PVDF). However, PVDF, as a fluorinated polymer and so-called â forever chemical,â requires complex synthesis under high-temperature and high-pressure conditions. Achieving functional polarization also involves another high-temperature, high-voltage poling step. The semi-crystalline morphology of PVDF can be tailored through chemical modificationsâ most notably by copolymerizationâ to endow it with intrinsically adjustable electromechanical properties without the need for electrical poling. For example, one may evolve from simple P(VDF-co-trifluoroethylene) [P(VDF-TrFE)]â based copolymers to more complex terpolymers and even tetrapolymers. However, in practice this route has proven infeasible for large-scale manufacture owing to its synthetic complexity and the attendant high production costs. Moreover, fabricating PVDF into ultra-thin films is challenging: thick films are incompatible with microelectrode architectures, while thin films are prone to dielectric breakdown during poling.

To address these limitations, this thesis explores polymer brush architectures as an alternative platform for energy conversion. Through surface-initiated polymerization, polymer chains are grafted at one end and extend in an oriented "brush" conformation. This backbone alignment, in turn, compels the pendant polar moieties to adopt, more or less, the same orientation, thereby generating an intrinsic polarization in the as-grafted thin film without the need for any post-treatments such as electrical poling. This architecture provides intrinsic chain ordering, controllable thickness, and excellent conformality, making it highly compatible with micro- and nanoscale device integration.

This thesis demonstrates standard â textbook-qualityâ pyroelectric responses in polar-functionalized polymer brushes, and confirms that the observed behavior originates from fixed dipole moments in the chain architecture. The results further reveal that the pyroelectric performance primarily arises from the dense, brush-like regions of the film where the chains are highly stretched and aligned. In contrast, in thicker brushes, the upper segments tend to adopt disordered coil conformations that contribute negligibly to the overall polarization, effectively diluting the polarision density.

These findings establish polymer brushes as a tunable platform for pyroelectric energy conversion. Beyond simplifying processing requirements, they also offer a model system for probing structureâ polarization relationships and pave the way for flexible, conformal, and energy-autonomous interfaces in next-generation microelectronic systems.